As a new lifestyle, face recognition has begun to be rapidly popularized in environments such as traveling by car, checking in, paying, applying for certificates, and public security and justice.
Google search for "face data" keywords, you will see many related public download indexes.
These data sets are usually used to train deep learning models of human faces. Although they are helpful for technological progress, they naturally also bring a series of new risks.
In February 2019, a large-scale data breach occurred in Shenzhen Deepnet Vision Technology. Its face recognition database contains more than 2.5 million personal information.
This information includes ID number, gender, address, birthday, pass, employer, and even the location of the camera they passed by in the past 24 hours.
Such as "hotel", "police", "Internet cafe", "restaurant", etc. related GPS location description.
data shows that in the past 24 hours alone, this company has collected as many as 6.68 million data.
In addition, the company’s face recognition database directly “runs naked” on the public network without security certification.
In other words, anyone can view these records and track one's actions.
Just a few days ago, the "AI face-changing" application ZAO made a lot of noise on the Internet, making everyone once again mention the importance of privacy.
Collecting face data without permission, infringing on citizens’ right to portrait, privacy and personal information security, is a crime.
But in the face of money, there are still people who take the risk of crime to collect and sell face data.
On September 11, Beijing Youth Daily published a piece of news, "The face data of 170,000 people has been publicly sold, and the parties have no knowledge of this."
According to reports, a reporter from the Beijing Youth Daily found in an online shopping mall that a business had publicly peddled face data, the number of which was about 170,000.
These data cover the portraits of two thousand people, each with about 50 to 100 photos.
Thinking more carefully, each photo is also matched with the corresponding data file. In addition to the location information, there are 106 key points of the face, such as eyes, The contours of ears, nose, mouth, eyebrows, etc.
In addition, the data also provides information such as the character's gender, expression emotion, facial value, and whether to wear glasses.
A reporter from Beiqing Daily bought the data package as a buyer and found that it did include 2000 folders. In addition to celebrities,
also contains photos from ordinary citizens such as doctors and teachers, including information on minors.
Subsequently, the reporter contacted multiple parties in the data packet, and they all said they knew nothing about it.
After a reporter's complaint, the product was removed from the shelves on September 9. But according to the merchant, he has sold these data many times since the release.
The merchant also said that he is also engaged in the artificial intelligence industry, and usually needs to collect this data, and sell it to earn money by the way.
comments from netizens: "Make money in prison."
Your facial data was collected without knowing it, and it became a profit-making tool for others , Just thinking about it is chilling.
Another thing to note is that this topic appeared on Weibo hot search on September 11. But as of September 12, Wuhen was posted on Weibotrace.
And you can also see this news in @ Sina's Weibo.
After clicking the tag of 170,000 people's face data that was publicly sold, it showed no content.
However, the news can still be found in the Qilu Evening News, Beijing News and other media electronic newspapers.
Technology is a double-edged sword. This kind of thing will inevitably happen, and it will happen all the time. It may only be the tip of the iceberg.
While enjoying the extreme convenience it brings, the security of personal information may also appear extremely fragile.
This is a normal law of industrial development, but it is a responsibility that various industries and regulatory agencies cannot ignore
